Converting XY to ORTF during post?

[jnorman]

hello boys and girls - i have a quick question for you audio experts.

ORTF presents a fairly clean stereo image with good locational information, based on a combination of intensity differences and temporal differences. spaced omnis present a less focused image since they only offer temporal differences between the Land R signals. XY is often criticized for having too narrow a soundstage image, in comparison with ORTF, since XY does not capture any of the temporal differences between the Land R signals.

so, my question - since the mechanical difference between XY configuration and ORTF, is simply the spacing between the mics (no spacing between XY, and approx 7" spacing for ORTF), and since that spacing is what generates the temporal information which helps create a more convincing soundstage, and since that temporal difference is effectively the amount of time delay between when a signal is captured by the L mic and the time at which the same signal is captured by the R mic:

can we convert an XY recording to something resembling an ORTF recording simply by delaying either the L or R track by the amount of time it takes sound to travel 7"?

similarly, can we, during post, effectively change the spacing between a pair of AB omnis, by the same method - simply delaying either the L or R track by a certain amount?

hope the question is clear. thanks for your comments and thoughts on this.

[krs]

Interesting idea, never thought of it. ORTF is more off axis - somewhat dependent on how wide the source is and your distance from it. Curious to hear what you come up with.

[fafalio]

go to the thread i have started about ambisonic for stereo. in that place you can find your answer.
with the harpex plug in or similar and an ambisonic mic you can change in post all you want. also the distance between mics.
but as you can read the ambisonics technology of these years seems not so good for stereo as a true stereo classical recording. at least in the theory.
on the other hand, now the more natural method to do what you want is to use ambisonic technology, because it records the 3d spatial information. with a stereo classical you record in 2d so if you try a plug in, it will always not so good as doing this in ambisonics. because it has recorded less spatial information.
Alessandro.

[apple-q]

Originally Posted by jnorman

similarly, can we, during post, effectively change the spacing between a pair of AB omnis, by the same method - simply delaying either the L or R track by a certain amount?

hope the question is clear. thanks for your comments and thoughts on this.

try it. pull a stereo recording into your DAW and nudge one of the tracks around by a couple of ms. How does it sound?
imagine having a singer in the middle front of your stereo mic. you singer appears in the phantom center of your stereo image. no matter how wide the distance between the mics, your singer will always be in the center of the stereo image within a certain range. applying delay to one of the mics does something completely different because you are not changing the physical position of the mics.
if what you propose worked, you could create a stereo image by using just one mic, splitting that and then apply a delay to one of the split signals.

back to my example: the time difference between the left and right mic relative to the sound source is zero. by applying a delay could within certain boudaries move the voice more to the left and right but that will not change the stereo base width at all.

[leddy]

Your brain uses time and intensity cues to localize sound. If you delay the right or left channel, you will force all the time cues to the earlier side. This will certainly conflict with some of the intensity cues and become a phasey, comb-filtered mess.

If you have an XY recording that sounds too narrow, try messing around with the mid-side ratio or adding some good reverb.

[d_fu]

Originally Posted by jnorman

can we convert an XY recording to something resembling an ORTF recording simply by delaying either the L or R track by the amount of time it takes sound to travel 7"?

Well, which track would you delay...? For signals from the extreme left, you'd need to delay L, and for signals from the extreme right, you'd need to delay R...
And what do you do with signals that are slightly off-center, or even dead center...?

D.

[leddy]

Originally Posted by d_fu

Well, which track would you delay...? For signals from the extreme left, you'd need to delay L, and for signals from the extreme right, you'd need to delay R...
And what do you do with signals that are slightly off-center, or even dead center...?

D.

I have played around with converting to MS, then delaying the side channels slightly. Sometimes that worked a little, but it was always a trade-off. Sounded a little more spacious at the expense of some comb-filtering, etc. In the end I gave up.

[Petrus]

Short answer: no.

Longer answer: if we apply a 7" time delay, that would work only for the signals coming straight from the sides. For the signals coming from straight ahead we need no delay at all. For signals coming from the angles between these extremes we need a delay from zero to 7" distance. There is no way to apply different delays for signals coming form different directions, as we can not tell which signal comes from where. At least there is no simple way.

There could be a way of measuring the intesities of similar signals and applying different delays depending on the intesity differencies (which should be related to the angular differencies, at least in a recordings done in an anechoic chamber). This requires a bit more than just delaying one track, if it is possible at all. Complicated DSP in any case and the end result woudl likely to be worse than the original.

[MichaelPatrick]

Petrus has it right. Temporal difference varies with angle of incidence, so there is no fixed time factor that can mimic an ORTF's vital, complex relation to instruments and acoustic.

[joelpatterson]

But... maybe if you delayed both tracks....

[boojum]

Originally Posted by joelpatterson

But... maybe if you delayed both tracks....

Net sum zero gain.

[mixedupsteve]

Izotope Ozone has a delay setting in the stereo mix function. It also has a variable mix amt. You might try it.

[jnorman]

thanks for all the responses - i just thought it might be an interesting idea, after reading all the information in the soundfield post and the way that software can simulate a variety of different configurations. i figured it was crazy or i would have already heard about it long ago.

boojum, please do not pay any attention to joelpatterson - he is clearly insane...

[Corran]

The speed of sound is about 340 meters per second. If you wanted a separation of 17cm (0.17 meters) the delay would be 0.17/340 seconds, or 0.5 milliseconds. So that should be your delay. (*fixed*)

The problem is, you can only delay one channel. Therefore, even if it worked, it would skew your stereo array to the L or R by 17cm. Might not matter, or maybe it does.

Secondly, the phase interactions between any stereo array are what makes the stereo sound - not just the temporal difference. So here is a conjecture - what if you changed the phase too? For example - if your XY setup was 3.4 meters from the sound source (to make the math easy!), that would be 5% change in phase. I am certain I have seen a plugin that will change the phase in increments, not just full one way or the other, so it might be cool to try? (*edit: Actually this would change the phase like the mic was 17cm behind the other mic. my bad.*)

[loujudson]

Originally Posted by joelpatterson

But... maybe if you delayed both tracks....

Yeah that's the ticket! Delay them both away from center! Ooh baby what stereo! NOT! Very amusing. NOT!

In my book there is no substitute for good mic placement, plugin or not.

But Joel is a cool guy, he is pulling your chains... both ways!

L

[joelpatterson]

Stop me before I post again!

[MichaelPatrick]

Try a Haas stereo effect. Lots of spacial plugins out there that can do variations on the Haas theme.

[PoxyMusic]

Originally Posted by joelpatterson

But... maybe if you delayed both tracks....

No, you need to move the left track forward in time 2.5 ms, while delaying the right track 2.5 ms.

Duh!

[RobAnderson]

What is proposed is technically impossible, however, I think Joel comes the closest to the correct way to try by saying you need to delay both channels.

There is one missing ingredient to his proposal: you would need to cross pan the delayed signals with the original XY recording.

In other words, you would have to delay the right channel and then pan this delayed signal to the left while keeping the original signal panned right. You would also have to delay the left channel and then pan this delayed signal to the right while keeping the original signal panned left.

I can only imagine this having a chance at working if you mixed the delayed signals at a very low level. Even then, you have not created an ORTF out of XY, but added some delay effect to create non-arbitrary phase differences between the channels.

[loujudson]

Rob, you may be on to something. Would be interesting to try duplicating the stereo track, make one mono at lower level, and delaying BOTH on the stereo track keeping them wide panned. Maybe it would sound okay?!?

Worth an experiment... sort of pseudo MS...

Quiet, Joel!

Lou

[boojum]

Originally Posted by RobAnderson

What is proposed is technically impossible, however, I think Joel comes the closest to the correct way to try by saying you need to delay both channels.

There is one missing ingredient to his proposal: you would need to cross pan the delayed signals with the original XY recording.

In other words, you would have to delay the right channel and then pan this delayed signal to the left while keeping the original signal panned right. You would also have to delay the left channel and then pan this delayed signal to the right while keeping the original signal panned left.

I can only imagine this having a chance at working if you mixed the delayed signals at a very low level. Even then, you have not created an ORTF out of XY, but added some delay effect to create non-arbitrary phase differences between the channels.

This is very like what Carver has done with his Sonic Hologram TM which did not work as well as the Hughes SRS. They screwed around with timing and channel swapping like this to expand the stereo image, and succeeded. I have both the black boxes. The SRS could spread a stereo image across a whole wall, way outside of the speakers. Magic.

[joelpatterson]

Of course... the obvious disadvantage with delaying the whole thing is that you'd have to sit around, waiting, to listen to it....

[Don S]

Originally Posted by RobAnderson

There is one missing ingredient to his proposal: you would need to cross pan the delayed signals with the original XY recording.

That's the closest I've heard. You would have to record the effect of the 7 inch time differential and (successfully) mix it with the properly delayed X-Y channels. I suspect the result would not sound convincing enough.

[MTStudios]

Guys, the time difference isnt just about the instruments, it's about the room. In an ORTF config, the instruments and room ambience are less blurred (thus more 3d) because the temporal information encoded in the delays between the mics is decoded by our brain in a very complex fashion. However, to put it simply, delaying anything or creating a faux 3d HAS to affect the reverb/early reflections in tandem with the instruments timbre. Once you've recorded it in XY, that information simply does not exist.

[d_fu]

I'm a bit surprised that this is still being discussed in all seriousness...

Originally Posted by Don S

You would have to record the effect of the 7 inch time differential

There is no such "differential" - at least only in theory for signals from the extreme left or right of the microphone, which is actually outside the recording angle of ORTF. And everything else is different in terms of delay...

[Tom McC]

Originally Posted by jnorman

thanks for all the responses - i just thought it might be an interesting idea, after reading all the information in the soundfield post and the way that software can simulate a variety of different configurations.

As Petrus and others already pointed out, you can't transform a, say, L-pointing coincident mic into a L-pointing ORTF mic for the simple reason that the delay required will vary with the direction of the arriving sounds, and the single mic capsule is essentially "blind"- it has no way of discerning from which direction the sounds are coming. So a blanket delay won't cut it.

With the Soundfield, however, one could expect more powerful transformations to be feasible because the tetrahedral array of 4 mic capsules at least supplies much more 3-D info for subsequent mathematical manipulations.

[RobAnderson]

Originally Posted by loujudson

Rob, you may be on to something. Would be interesting to try duplicating the stereo track, make one mono at lower level, and delaying BOTH on the stereo track keeping them wide panned. Maybe it would sound okay?!?

Worth an experiment... sort of pseudo MS...

Quiet, Joel!

Lou

MS would actually work better in this case I think, if the desired result is a weaker center and more width. I wrote a short paper on how to do this if you don't have a plug-in - it's on my website:

Rob Anderson's Online Portfolio

Just scroll down to the "Selected Writings"

[mpdonahue]

Originally Posted by Corran

The speed of sound is about 340 meters per second. If you wanted a separation of 17cm (0.17 meters) the delay would be 17/340 seconds, or 1/20 of a second or 5 milliseconds. So that should be your delay.

Just FYI, you are missing a decimal place in your math.....

All the best,
-mark

[rumleymusic]

Just FYI, you are missing a decimal place in your math.....

Yeah, 5ms is about 5.5- 6 feet in delay.

Spreading out the image with delay is possible without shifting the image right or left, but it needs to be done in an MS codec with reference to center. Sonar's Channel Tools does exactly that, and it works quite well. I don't think anything will change XY into ORTF though, it is and entirely different angle with unrelated off axis pickup.

[Corran]

Whoops, my bad. I was just banging that out and didn't double-check. (fixed it)

Also I didn't properly apply the right trig for the phase difference.

Whatever - anyway, of course it isn't truly possible but you might get something interesting, even if it wasn't truly a transformation of XY to ORTF.